Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import torch
|
3 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
4 |
+
|
5 |
+
def run_LLM (model, tokenizer, streamer, prompt):
|
6 |
+
|
7 |
+
token_ids = tokenizer.encode(prompt, return_tensors="pt")
|
8 |
+
output_ids = model.generate(
|
9 |
+
input_ids=token_ids.to(model.device),
|
10 |
+
#max_new_tokens=300,
|
11 |
+
max_new_tokens=3000000,
|
12 |
+
do_sample=True,
|
13 |
+
temperature=0.8,
|
14 |
+
)
|
15 |
+
|
16 |
+
n_tokens = len(output_ids[0])
|
17 |
+
output_text = tokenizer.decode(output_ids[0])
|
18 |
+
|
19 |
+
return (output_text, n_tokens)
|
20 |
+
|
21 |
+
def display_message():
|
22 |
+
model = AutoModelForCausalLM.from_pretrained("cyberagent/calm2-7b-chat",
|
23 |
+
device_map="cuda",
|
24 |
+
torch_dtype="auto")
|
25 |
+
tokenizer = AutoTokenizer.from_pretrained("cyberagent/calm2-7b-chat")
|
26 |
+
streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
|
27 |
+
|
28 |
+
prompt = """わが国の経済について今後の予想を教えてください。
|
29 |
+
ASSISTANT: """
|
30 |
+
|
31 |
+
(result, n_tokens) = run_LLM(model, tokenizer, streamer, prompt)
|
32 |
+
|
33 |
+
return result
|
34 |
+
|
35 |
+
|
36 |
+
if __name__ == '__main__':
|
37 |
+
|
38 |
+
iface = gr.Interface(fn=display_message, inputs=None, outputs="text")
|
39 |
+
iface.launch()
|